classiﬁcation of human chromosome 21 gene-expression ... · down syndrome caused by chromosome 21...

www.ajhg.org The American Journal of Human Genetics Volume 81 September 2007 475

ARTICLE

Classification of Human Chromosome 21 Gene-ExpressionVariations in Down Syndrome: Impact on Disease PhenotypesE. Aı̈t Yahya-Graison, J. Aubert, L. Dauphinot, I. Rivals, M. Prieur, G. Golfier, J. Rossier,L. Personnaz, N. Créau, H. Bléhaut, S. Robin, J. M. Delabar, and M.-C. Potier

Down syndrome caused by chromosome 21 trisomy is the most common genetic cause of mental retardation in humans.Disruption of the phenotype is thought to be the result of gene-dosage imbalance. Variations in chromosome 21 geneexpression in Down syndrome were analyzed in lymphoblastoid cells derived from patients and control individuals. Ofthe 359 genes and predictions displayed on a specifically designed high-content chromosome 21 microarray, one-thirdwere expressed in lymphoblastoid cells. We performed a mixed-model analysis of variance to find genes that are differ-entially expressed in Down syndrome independent of sex and interindividual variations. In addition, we identified geneswith variations between Down syndrome and control samples that were significantly different from the gene-dosageeffect (1.5). Microarray data were validated by quantitative polymerase chain reaction. We found that 29% of the expressedchromosome 21 transcripts are overexpressed in Down syndrome and correspond to either genes or open reading frames.Among these, 22% are increased proportional to the gene-dosage effect, and 7% are amplified. The other 71% of expressedsequences are either compensated (56%, with a large proportion of predicted genes and antisense transcripts) or highlyvariable among individuals (15%). Thus, most of the chromosome 21 transcripts are compensated for the gene-dosageeffect. Overexpressed genes are likely to be involved in the Down syndrome phenotype, in contrast to the compensatedgenes. Highly variable genes could account for phenotypic variations observed in patients. Finally, we show that alter-native transcripts belonging to the same gene are similarly regulated in Down syndrome but sense and antisense transcriptsare not.

From the Neurobiologie et Diversité Cellulaire, Unité Mixte de Recherche 7637 du Centre National de la Recherche Scientifique et de l’Ecole Supérieurede Physique et de Chimie Industrielles de la Ville de Paris (E.A.Y.-G.; L.D.; G.G.; J.R.; M.-C.P.), Université Paris Diderot-Paris (E.A.Y.-G.; N.C.; J.M.D.),Unité Mixte de Recherche de l’Ecole Nationale du Génie Rural, des Eaux et des Forêts, de l’Institut National Agronomique Paris-Grignon et de l’InstitutNational de la Recherche Agronomique (J.A.; S.R.), Equipe de Statistique Appliquée (I.R.; L.P.), Service de Cytogénétique, Hôpital Necker Enfants Malades(M.P.), and Institut Jérôme Lejeune (H.B.), Paris

Received March 13, 2007; accepted for publication May 21, 2007; electronically published July 19, 2007.Address for correspondence and reprints: Dr. M.-C. Potier, Neurobiologie et Diversité Cellulaire, CNRS UMR 7637, ESPCI, 10 rue Vauquelin, 75005

Paris, France. E-mail: [email protected]. J. Hum. Genet. 2007;81:475–491. � 2007 by The American Society of Human Genetics. All rights reserved. 0002-9297/2007/8103-0006$15.00DOI: 10.1086/520000

Down syndrome (DS [MIM #190685]) results from thetriplication of chromosome 21 and is the most commongenetic cause of mental retardation in humans, occurringin ∼1 in 800 newborns. The phenotype of DS is charac-terized by 180 clinical features, including cognitive im-pairments, muscle hypotonia, short stature, facial dys-morphisms, congenital heart disease, and several otheranomalies.1 These clinical features can vary considerablyin number and in severity,2 and certain abnormalities,such as acute megakaryoblastic leukemia and Hirsch-sprung disease, occur at higher frequencies in patientswith DS than in the general population.

Trisomy 21 has been known to be the cause of DS since1959, when Lejeune and colleagues demonstrated thepresence in three copies of chromosome 21 in personswith DS.3 The phenotype of DS is thus thought to be theresult of gene-dosage imbalance. However, the molecularmechanisms by which such dosage imbalance causes ab-normalities remain poorly understood. Two different hy-potheses have been proposed to explain the phenotypeof DS: “developmental instability” (loss of chromosomalbalance) and “gene-dosage effect.” According to the de-velopmental instability hypothesis, the presence of a su-

pernumerary chromosome globally disturbs the correctbalance of gene expression in DS cells during develop-ment.4,5 However, this hypothesis is weakened by the factthat other autosomal trisomy syndromes do not lead tothe same clinical pattern.6 Moreover, correlations betweengenotype and phenotype in patients with partial trisomiesindicate that a restricted region in 21q22.2 is associatedwith the main features of DS, including hypotonia, shortstature, facial dysmorphies, and mental retardation.7–9

This DS chromosomal region (DCR) supports the alter-native gene dosage–effect hypothesis, which postulatesthat the restricted number of genes from chromosome 21that are overexpressed in patients with segmental triso-mies contributes to the phenotypic abnormalities.

To determine which hypothesis applies to the etiologyof DS, several gene-expression studies of human DS cellsor tissues have been conducted.10–17 Most of these studieshave shown a global up-regulation of the three-copy genesmapping to the trisomic chromosome, but the limitednumber of studied DS cases restricted the statistical anal-ysis and did not allow the identification of precise genederegulation. Moreover, these studies were performed us-ing a small number of three-copy genes. Several other ex-

476 The American Journal of Human Genetics Volume 81 September 2007 www.ajhg.org

Table 1. Experimental Design

Controls

Men with DS Women with DS

1 2 3 4 5 6 7 8 9 10

Men:11 �1 1 �1 1 �1 112 1 �1 1 �113 �1 1 �1 1 1 �114 1 �1 �1 1

Women:15 1 �116 �1 117 1 �1 1 �118 �1 1 �1 119 1 �120 �1 121 �1 1 �1 1

NOTE.—Microarray experiments were performed using LCLs from indi-viduals with DS and control individuals in accordance with a mixed model(see the “Material and Methods” section). Each “1” indicates one ex-periment. “�1” means that DS and control samples were labeled withCy5 and Cy3, respectively. “�1” means that DS and control samples werelabeled with Cy3 and Cy5, respectively.

Table 2. List of Oligonucleotide PrimersUsed in the QPCR Experiments

PrimerSequence(5′r3′)

CHAF1B_UP CCATCATATGGGATGTCAGCAACHAF1B_LOW CTTCATGCTGTCGTCGTGAAACCSTB_UP GCCACCGCCGAGACCCAGCACSTB_LOW TGGCTTTGTTGGTCTGGTAGDSCR1_UP GCACAAGGACATTTGGGACTDSCR1_LOW TTGCTGCTGTTTTCACAACCDYRK1A_UP ATCCGACGCACCAGCATCDYRK1A_LOW AATTGTAGACCCTTGGCCTGGTGART_UP CTGGGATTGTTGGGAACCTGAGGART_LOW ACCAAAGCAGGGAAGTCTGCACH2BFS_UP CAGAAGAAGGACGGCAGGAAH2BFS_LOW GAAGCCTCACCTGCGATGCGMX1_UP GCCAGTATGAGGAGAAGGTGCGMX1_LOW GTTTCAGCACCAGCGGGCATCTSNF1LK_UP GCCGCTTCCGCATCCCCTTCTTSNF1LK_LOW CTCATCGTAGTCGCCCAGGTTGSOD1_long_UP TCGCCCAATAAACATTCCCTTGSOD1_long_LOW AAGTCTGGCAAAATACAGGTCATTGSTCH_UP GGACGTGGCCTTTCTGATAASTCH_LOW CTTGACGGATCCGAGGAATATMEM1_UP CGTGCAGGAACTGAAGCTCTTATMEM1_LOW TCTGAGCTGTGTTGGCTGTTTCL13852_UP CTCCAATCTCAGCCGTCAGTL13852_LOW AGCCACACCATCCACACGGGAB000468_UP CAAGAAAGCGTCGTGGTGGAAB000468_LOW ATCGTCACTGCTCACCACAC

periments have been done on animal models of DS witha greater number of chromosome 21 gene orthologs byuse of microarray and quantitative PCR experiments.18–21

In these studies, the three-copy genes were overexpressed,with a mean ratio of 1.5, which is proportional to thegene-dosage imbalance. However, some of these tripli-cated genes appeared to escape the “1.5-fold rule.” Yet,these animal models are not trisomic for all chromosome21 orthologs. Thus, a comprehensive classification of allhuman genes on chromosome 21, according to their levelof expression in DS, does not yet exist.

The goal of the present study was to fill this knowledgegap and to find the genes that are likely to be involvedin DS phenotypes through their transcriptional dysregu-lation.22 For this purpose, we designed an oligonucleotidemicroarray containing all chromosome 21 genes, ORFs,antisense transcripts, and predicted genes listed in themost common databases (NCBI Gene Database, EleanorRoosevelt Institute, and Max Planck Institute), except forthe 53 genes of the keratin-associated protein cluster. Geneexpression was measured on lymphoblastoid cell lines(LCLs) from 10 patients with DS and 11 control individ-uals. LCLs are easy to obtain and are widely used to studygenotype-phenotype correlations.23 To our knowledge,this is the most comprehensive study so far that has beendone using triplicated genes in DS human cells. In addi-tion, we analyzed data with a mixed-model analysis ofvariance, to find genes that are differentially expressed inDS independent of sex and interindividual variations. Ourdata show a global gene dosage–dependent expression ofchromosome 21 genes in LCLs, with no effect of sex. Inaddition, by use of our data-analysis protocol, chromo-some 21 genes can now be classified into four classes: classI genes are overexpressed with a mean ratio very close to1.5, proportional to the gene-dosage effect of trisomy 21;

class II genes are overexpressed with ratios significantly11.5, reflecting an amplification mechanism; class IIIgenes have ratios significantly !1.5, corresponding tocompensated genes; and class IV genes have expressionlevels that are highly variable between individuals. Thisclassification should have an impact on the search forgenes that are involved in the DS phenotype.

Material and MethodsCell Lines and Culture Conditions

LCLs were derived from the B lymphocytes of 10 patients withDS collected from the cytogenetic service of the hospital NeckerEnfants Malades and the Institut Jérôme Lejeune. Parents of pa-tients from the Institut Jérôme Lejeune gave their informed con-sent, and the French biomedical ethics committee gave its ap-proval for this study (Comité de Protection des Personnes dansla Recherche Biomédicale number 03025). Written informed con-sent was obtained from the participants or from their families bythe cytogenetic service of Hôpital Necker Enfants Malades. Celllines from 11 control individuals were also obtained with theirwritten informed consent, for comparison of chromosome 21gene-expression profiles. Culture media consisted of Opti-MEMwith GlutaMax (Invitrogen) supplemented with 5% fetal bovineserum from a unique batch and 1% penicillin and streptomycinmix (10,000 U/ml). Cell lines were grown at 37�C in humidifiedincubators, in an atmosphere of 5% CO2. Each culture was grownto at least cells. All cell lines were karyotyped, to confirm660 # 10their trisomic or euploid status and also to verify that immor-talization by the Epstein-Barr virus (EBV) did not produce anyvisible chromosomal rearrangement other than trisomy 21. Cells


Table 3. HSA21 Oligoarray Content

Putative ExpressedSequences HSA21 Contenta

HSA21 Oligoarray Content

No. ofSequencesb

HSA21Coveragec

(%)

Genes 182 145 82.42ORFs 93 58 62.37Predictions Not represented 118d NAAntisense transcripts Only 1 represented 18 NA

a NCBI Gene Database build 36.2 was used to estimate HSA21 gene content. Onlycurrent sequences were considered, with the exception of pseudogenes and hypo-thetical proteins.

b The number of HSA21 sequences represented by at least one probe on the HSA21oligoarray.

c The percentage of HSA21 sequences currently annotated in NCBI Gene Databasethat are represented on the microarray. NA p not available.

d Of the 118, 20 are represented with their reverse sequence.

Table 4. Classification of HSA21 Genes by StatisticalAnalysis

DS/Control Ratio

Class by DS/Control Ratio

Not SignificantlyDifferent from 1

SignificantlyDifferent from 1

Significantly 11.5 … IINot significantly

different from 1.5 IV ISignificantly !1.5 III …

were harvested by centrifugation, were washed in 5 ml PBS, fol-lowed by another centrifugation, and were stored at �80�C.

Human Chromosome 21 (HSA21) Oligoarray

A dedicated oligonucleotide microarray—named “HSA21 oli-goarray,” containing 664 50-mer amino-modified oligonucleo-tides representing 145 genes, 58 ORFs, 118 predictions (20 ofthem represented in both orientations), and 18 antisense tran-scripts assigned to chromosome 21—was used in the presentstudy. Predictions represented on the array included cDNAs andexons from the CBR-ERG region on 21q deduced from cDNA iso-lation and exon-trapping experiments8 and gene or exon predic-tions produced from in silico analysis of the complete sequenceof human chromosome 21.24 Nonredundant transcript sequencesand antisense transcripts were also included in this oligoarray.25–27 Thirty-nine genes assigned to chromosomes other than chro-mosome 21, represented by 58 oligonucleotides—showing a widerange of expression levels according to UniGene and no varia-tions between DS and control samples as demonstrated by thefirst version of the HSA21 oligoarray (data not shown)—wereadded for data normalization. All probes present on the arraywere designed using the SOL software (G.G., S. Lemoine, A. Bend-joudi, J.R., S. Lecrom, and M.-C.P., unpublished data). Sequenceswere then synthesized by EuroGentec and were spotted ontoCodeLink activated glass slides (Amersham Biosciences) by use ofa MicroGrid II spotter (Biorobotics). Each array contained twomatrices with eight blocks each, in which probes were present induplicates so that each oligonucleotide was present in four rep-licates on each slide.

Experimental Design

The experiment comprised 10 patients with DS (7 men and 3women) and 11 controls (4 men and 7 women). Samples fromthe same individual were used in different hybridizations.

For each gene, we used the linear model

y p m � D � S � A � F � DSijklm i j l m ij

�DF � SF � I(DS) � � , (1)im jm ijk ijklm

where is the normalized expression of the gene in log2 foryijklmfactor i (DS or control sample), sex is j, patient number is k( ), dye label is m (m is red, for Cy5, or green, for Cy3)k p 1, … ,21on the HSA21 oligoarray l. The symbols D, S, A, and F representthe fixed effects due to the disease, sex, array, and fluorochrome,respectively. For example, D represents the modifications of thegene expression level due to the disease. A and F are both nuisanceparameters that account for potential technological biases. DS,DF, and SF correspond to interacting effects: disease and sex, dis-ease and fluorochrome, and sex and fluorochrome, respectively.The symbol I(DS) refers to the patient (nested within disease andsex) random effect. This last effect accounted for the correlationbetween samples used in different hybridizations but collectedfrom the same patient.

We assumed the independence between all I(DS)ijk and Eijklm. Wealso assumed that I(DS)ijk was independent with a distributionN(0,s2) and that Eijklm was independent with distribution N(0,j

2).Model (1) can be rewritten under the matrix form

Y p Xv � ZU � E , (2)

where v is the vector of fixed effects (D, S, A, F, DS, DF, and SF),U is the vector of I(DS)ijklm, and E is the vector of Eijklm. Y ∼

, where . Y has n rows (n is the total2 2 TN(Xv,S) S p 2j Id � s ZZg gnumber of samples) and one column, X is the matrix describingthe status of the patient (disease and sex) from which the samplewas collected. Z has n rows and I columns (I is the total numberof patients) and describes the correspondence between samplesand patients.


Figure 1. Classification of HSA21 genes according to the expression ratio between DS and control LCLs. The sum of classified genesis 136 genes minus 2 (C21ORF108 and PRMT2) that appear twice, depending on the oligonucleotide probe considered (see the “Results”section for details).

On each array l, we actually observed the differential expression(red signal minus green signal)

y p y � y p (D � D ) � (S � S )′ ′ ′ ′ ′ ′ ′ ′ ′ ′ii jj kk lmm ijklm i j k lm i i j j

�(DS � DS ) � (F � F )′ ′ ′ij i j m m

�(DF � DF ) � (SF � SF )′ ′ ′ ′im i m im i m

�[I(DS) � I(DS) ] � (� � � ) . (3)′ ′ ′ ′ ′ ′ ′ ′ijk i j k ijklm i j k l m

This model can be rewritten under another matrix, D, describingthe comparisons performed on each array. This matrix had L rows(L is the total number of arrays) and n columns. The lth row ofD was zero except for the value 1 in the column correspondingto the sample labeled in red and �1 in the column correspondingto the sample labeled in green. The model for the differentialexpression was obtained by multiplying all terms of equation (2)by D:

DY p DXv � DZU � DE .

The vector of differential expression DY has distributionN(DXv,DSDT).

The experimental design was defined by the three matrices X,Z, and D. X and Z basically depend on the number of samplesfor each patient. Because the microarrays used were two-colorassays, the total number of samples was twice the number ofslides.

The important remaining choice was the comparison to bemade—that is, the choice of D. Consideration of the differentialexpression between two patients analyzed on the same array elim-inated the array effect and the corresponding constants. We werenot interested in other technical effects, such as fluorochrome orinteractions of fluorochrome with other effects. To eliminate DFimand SFjm effects, we proposed a balanced design for the fluoro-

chrome effect. Finally, data were normalized across genes, to re-move the dye effect .Fm

The last criterion was D—the precision of the estimated effects(gathered into the vector ). According to the mixed linear modelv̂theory, this precision is given by its variance matrix,

T T T �1 �1ˆV(v ) p [D X (DSD ) DX] ,A

which depends on D and the ratio j2/s2. The diagonal containedall the information about the quality of the estimates. It gave thevariance of the estimates of all effects of interest to , , andD Si j

. This matrix was the ultimate tool for comparing the designs.DSijWe calculated the variance of the disease effect (which is of

primary interest) and the determinant of the variance covariancematrix (which gave a global measure of the precision of the es-timates) for a certain number of designs D. To do that, we useda value given a priori for the ratio j2/s2 equal to 2.

We finally chose a design involving 40 arrays. Each patientappeared in two to eight different experiments, and samples fromthe same patient were marked the same number of times witheach fluorescent dye (Cy3 or Cy5). On each array, a DS LCL anda control LCL were compared, to increase the precision of thedisease effect. Ten arrays compared (i) a man with DS and anunaffected man, (ii) a female with DS and an unaffected female,(iii) a man with DS and an unaffected female, or (iv) a femalewith DS and an unaffected male. The design is described in table1.

mRNA Extraction, HSA21 Oligoarray Hybridization, DataFiltering, and Normalization

mRNA was extracted from frozen individual cell samples by useof Fast Track 2.0 mRNA Isolation kit (Invitrogen) in accordancewith the manufacturer’s instructions. To eliminate DNA contam-ination, the appended DNase protocol of RNeasy mini kit (Qia-


gen) was used in accordance with the manufacturer’s protocol.Samples were further tested for purity and quantity with RNA6000 NanoChips by use of the Agilent 2100 Bioanalyzer (AgilentTechnologies). By use of the Reverse-iT RTase Blend kit (ABGene),2 mg of mRNA was converted into Cy3- or Cy5-labeled cDNA byincorporation of fluorescent dUTP (Amersham). Labeled targetswere then purified on Qiaquick columns in accordance with themanufacturer’s protocol (Qiagen). Hybridization of sample pairson HSA21 oligoarrays (one DS sample and one control sample),according to the experimental design, was performed using hy-bridization buffer (50% formamide, 4# saline sodium citrate[SSC], 0.1% SDS, and 5# Denhart) at 42�C overnight. Slides werewashed in 2# SSC and 0.1% SDS three times for 5 min, in 0.2#SSC for 1 min, and in 0.1# SSC for 2 min. Data were acquiredwith GenePix 4000B scanner and by use of the GenePix Pro 6.0software (Axon). For each array, the raw data comprised the me-dian feature pixel intensity at wavelengths 635 nm and 532 nmfor Cy5 and Cy3 labeling, respectively. After subtraction of thebackground signal, LOWESS normalization28 of the M values cor-responding to Cy5/Cy3 signal ratios in log2 was applied to alloligonucleotides representing non-HSA21 genes and was used tocalculate a correction factor applied to M values for HSA21 probesunder The R Project for Statistical Computing. Normalized datafrom each slide was then filtered using two criteria: (i) for eacholigonucleotide, at least two values among the four replicates hadto be available, and (ii) SD of values corresponding to the geo-metric mean in log2 of Cy3 and Cy5 signal intensities (the A value)had to be !1. Arithmetic means of normalized and filtered M andA values were calculated for each oligonucleotide and were sub-mitted to the statistical analysis. All microarray data used in thisstudy were deposited in the Gene Expression Omnibus (GEO)database (accession number GSE6408).

Expression Data Analysis: Statistical Testing

Mixed model.—To determine differentially expressed genes forDS, sex, and DS # sex effects, we performed a mixed-model anal-ysis of variance according to the experimental design. Thismethod was chosen to distinguish between interindividual var-iability and experimental variability.29 We used the mixed pro-cedure of the SAS software with the restricted maximum likeli-hood (REML) method of estimation.30 After the filtering andnormalization steps, the number of observations per spot variedbetween 8 and 40, which was enough to calculate the variancefor each gene.

We first tested the effects of the complete model (3). Since thesex and DS # sex effects were not significant for any gene, thesetwo effects were dropped from the model. We finally analyzedthe simplified model

y p y � y′ ′ ′ ′ ′ ′ii jj kk l ijkl i j k l

p (D � D ) � [I(DS) � I(DS) ] � (� � � ) .′ ′ ′ ′ ′ ′ ′ ′ ′i i ijk i j k ijklm i j k l m

We deduced raw P values from comparison with 1 under thenull hypothesis and adjusted them by the Benjamini-Hochbergprocedure, which controls the false-discovery rate (FDR).31 Wethen analyzed the simplified model by comparison with 1.5 underthe null hypothesis and adjusted the raw P values by use of themethod described by Storey et al.32

Principal-components analysis (PCA).—Results from microarray ex-periments were obtained as the differential expression between

DS and control samples. M values corresponded to log (DS) �2, and A values to . Forlog (control) [log (DS)� log (control)]/22 2 2

each probe p, the mean value of its expression (in log2) in DS celllines and in controls could thus be expressed as

k1 Mpi kE p A � for i p 1–10 (DS)�p p( )N 2k�Sii ,k1 Mp{ }i kE p A � for i p 11–21 (controls)�p p( )N 2k�Siiwhere i denoted the index of the individual, Si the set of slideson which all samples from the individual i have been hybridized,and Ni the size of this set. Since the overall expression level ofthe genes of one individual varied from one slide to another andto avoid normalization across slides, we made the simplification

and reconstructed relative mean values of expression EkA p 0pfor each individual as

k1 MpiE p for i p 1–10 (DS)�p ( )N 2k�Sii .k1 Mp{ }iE p � for i p 11–21 (controls)�p ( )N 2k�SiiPCA of chromosome 21 genes and genes mapping to other

chromosomes was performed separately using E values deducedfrom all expressed chromosome 21 probes and all expressed non–chromosome 21 probes, respectively.

PCR Experiments

To validate expression ratios between DS and control samplesobtained from HSA21 oligoarray data, 100 ng of mRNA was re-verse transcribed into cDNA by use of Reverse-iT RTase Blend kit(ABgene). Quantitative real-time PCR (QPCR) on diluted cDNAwas conducted in the presence of 0.6 mM of each specific primer(designed by Primer3 software) and 1# Quantitect SYBR GreenPCR master mix (Qiagen) containing 2.5 mM MgCl2, Hotstart Taqpolymerase, dNTP mix, and the fluorescent dye SYBR Green I.QPCR experiments were performed in a Lightcycler system(Roche Molecular Biochemicals) on 11 HSA21 genes: CHAF1B,CSTB, DSCR1, DYRK1A, GART, H2BFS, MX1, SNF1LK, SOD1,STCH,and TMEM1. The ubiquitin-activating enzyme E1 (NCBI Entrezaccession number L13852) mRNA mapping to HSA3 and the zinc-finger protein (NCBI Entrez accession number AB000468) mRNAmapping to HSA4 were used as endogenous control genes, asdescribed by Janel et al.33 For each sample, the mean cycle thresh-old value, Ct, was corrected by subtracting the mean of the Ctobtained with the two reference genes. PCR primers are listed intable 2.

ResultsDesign of a Comprehensive HSA21 Oligoarray

The HSA21 oligoarray was designed for the exhaustivestudy of human chromosome 21 gene expression in DS.This microarray contained 664 sequences representing145 genes, 58 ORFs, 118 predictions (plus the reverse se-quences for 20 of them), and 18 antisense transcripts, al-lowing expression analysis of all putative genes mappingto chromosome 21 and related to DS. To increase the

480

Table 5. List of Genes Classified According to Their Expression in DS LCLs

Gene Symbol

GenBankAccessionNumber Ratio A M Var(M) Classa

as-TTC3 BF979681.2 1.56 9.07 .64 .64 IC21orf108 (exon 39) AF231919.1 1.30 7.84 .38 .18 IC21orf33 BI824121.1 1.52 10.54 .60 .20 IC21orf59 AF282851.1 1.35 9.61 .44 .21 IC21orf66 AY033903.1 1.51 9.50 .59 .12 IC21orf7 AY171599.2 1.43 8.00 .52 .49 IC21orf91 BC015468.2 1.59 9.50 .67 .26 ICBS AF042836.1 1.61 8.13 .69 .48 ICCT8 BC095470.1 1.65 12.64 .72 .32 ICRYZL1 BC033023.2 1.52 9.45 .60 .12 IDONSON AF232673.1 1.42 9.73 .50 .11 IDYRK1A D86550.1 1.41 10.20 .49 .17 IHMGN1 M21339.1 1.38 11.82 .46 .10 IIFNAR1 AY654286.1 1.47 8.10 .56 .18 IIFNAR2 BC013156.1 1.67 8.38 .74 .14 IIFNGR2 AY644470.1 1.45 10.35 .54 .24 IIL10RB BT009777.1 1.66 10.18 .73 .21 IITGB2 BC005861.2 1.61 11.34 .68 .40 IMCM3AP AY590469.1 1.45 11.69 .54 .19 IMRPL39 AF109357.1 1.47 9.82 .55 .15 IPFKL X15573.1 1.50 12.44 .58 .23 IPIGP AF216305.1 1.60 8.14 .68 .25 IPTTG1IP NM_004339.2 1.52 9.89 .60 .20 ISFRS15 AF023142.1 1.39 10.42 .47 .14 ISLC5A3 L38500.2 1.57 8.92 .66 .18 ISON AY026895.1 1.51 12.64 .60 .14 ISUMO3 BC008420.1 1.41 10.82 .50 .10 IUSP16 AY333928.1 1.46 9.94 .54 .10 IUSP25 AF170562.1 1.58 9.93 .66 .10 IZNF294 NM_015565.1 1.51 8.77 .60 .11 IBTG3 D64110.1 1.82 10.63 .86 .21 IIC21orf57 AY040875.1 1.74 9.49 .80 .28 IIMRPS6 AB049942.1 1.64 10.26 .72 .13 IIPDXK AY303972.1 1.71 10.06 .77 .23 IISAMSN1 AF222927.1 2.27 10.47 1.18 .72 IISLC37A1 AF311320.1 1.72 9.33 .78 .32 IISNF1LK AB047786.1 2.14 9.47 1.10 1.15 IISTCH U04735.1 1.97 9.86 .98 .42 IITTC3 D84296.1 1.79 10.59 .84 .20 IIaa071193 AA071193.1 .97 7.34 �.05 .48 IIIAIRE AB006682.1 .82 7.28 �.28 .29 IIIAL041783 AL041783.1 1.06 7.10 .09 .36 IIIas-C21orf56 BC084577.1 1.07 11.75 .10 .08 IIIas-KIAA0179 AA425659.1 1.16 8.08 .22 .67 IIIATP5J BC001178.1 1.35 7.99 .44 .11 IIIB184 AL109967.2 1.06 9.10 .08 .62 IIIB27 inverse AP000034.1 .83 8.40 �.27 .31 IIIC21orf108 (exon 26) AF231919.1 1.09 8.86 .13 .09 IIIC21orf12 AP001705.1 .97 7.49 �.05 .72 IIIC21orf2 NM_004928.1 1.30 8.80 .38 .14 IIIC21orf21 AA969880 1.16 7.33 .22 .28 IIIC21orf25 AB047784.1 1.19 8.61 .25 .31 IIIC21orf29 AJ487962.1 .98 7.27 �.03 .22 IIIC21orf34 AF486622.1 .93 8.17 �.11 .28 IIIC21orf42 AY035383.1 1.14 10.05 .19 .22 IIIC21orf45 AF387845.1 1.23 9.39 .30 .12 IIIC21orf49 BC117399.1 1.14 7.80 .19 .32 IIIC21orf51 AY081144.1 1.26 9.47 .33 .09 IIIC21orf54 AA934973.1 1.01 7.26 .01 .38 IIIC21orf58 BC028934.1 1.11 7.73 .15 .21 IIIC21orf6 BC017912.1 1.20 9.63 .26 .22 IIICHAF1B U20980.1 1.29 8.35 .37 .13 IIICLIC6 AF448438.1 .74 9.26 �.43 .98 III

(continued)

481

Table 5. (continued)

Gene Symbol


CXADR AF200465.1 .90 8.16 �.15 .32 IIID21S2056E U79775.1 1.33 10.55 .41 .18 IIIDCR1-17.0 AJ001875.1 1.13 7.16 .17 .71 IIIDCR1-19.0 AJ001906.1 .94 7.13 �.09 .19 IIIDCR1-20.0-reverse AJ001893.1 .95 7.14 �.07 .36 IIIDCR1-25.0-reverse AJ001905.1 .96 7.34 �.06 .31 IIIDCR1-7.0 AJ001861.1 1.09 7.48 .12 .27 IIIDCR1-7.0-reverse AJ001861.1 1.21 7.13 .28 .28 IIIDCR1-8.0 AJ001862.1 .98 7.58 �.03 .42 IIIDCR1-8.0-reverse AJ001862.1 1.00 8.61 .00 .20 IIIDSCAM_Intronic_Model BG221591.1 1.21 8.39 .28 1.39 IIIDSCR1 AY325903.1 .93 7.13 �.10 .74 IIIDSCR10 AB066291.1 .95 8.60 �.07 .25 IIIDSCR2 AY463963.1 1.25 10.98 .32 .18 IIIDSCR3 D87343.1 1.40 10.09 .48 .12 IIIDSCR6 AB037158.1 1.03 7.37 .04 .41 IIIDSCR9 BC066653.1 1.05 7.69 .07 .28 IIIETS2 J04102.1 1.40 7.18 .49 .43 IIIGABPA BC035031.2 1.40 8.88 .49 .08 IIIGART X54199.1 1.17 9.55 .23 .15 IIIH2BFS AB041017.1 1.13 13.19 .17 .57 IIIHLCS AB063285.1 1.32 8.48 .40 .18 IIIICOSLG AF289028.1 1.23 9.99 .30 .28 IIIJAM2 AY016009.1 .98 8.07 �.03 .82 IIIKCNE1 BC046224.1 1.12 7.18 .16 .40 IIIKIAA0179 D80001.1 1.21 10.15 .28 .24 IIIMORC3 BC094779.1 1.34 8.51 .42 .15 IIIn74695 N74695 .93 9.83 �.10 .18 IIIPKNOX1 AY196965.1 1.04 7.81 .05 .12 IIIPOFUT2 NM_015227.3 1.35 7.46 .43 .49 IIIPRED21 AP001693.1 .95 7.75 �.07 .66 IIIPRED24 AP001695.1 1.00 7.41 .00 .65 IIIPRED41 AP001726.1 .93 7.15 �.11 .64 IIIPRED59 AL163301.2 .94 7.94 �.09 .17 IIIPRED63 AP001759.1 .99 7.56 �.01 .18 IIIPRED65 AL163202.2 1.06 7.10 .08 .25 IIIPRMT2 (exon 5/6) U80213.1 1.34 8.67 .42 .14 IIIPWP2H U56085.1 1.34 9.43 .42 .11 IIIRUNX1 D43968.1 .85 7.91 �.23 .76 IIISETD4 AF391112.1 1.09 9.24 .13 .21 IIISH3BGR X93498.1 1.07 7.19 .10 .83 IIISLC19A1 AF004354.1 1.20 7.80 .26 .14 IIISOD1b AY835629.1 1.15 9.55 .21 .21 IIISYNJ1 AF009039.1 1.29 8.07 .37 .13 IIITFF3 BC017859.1 .97 8.00 �.04 .46 IIITMEM1 BC101728.1 1.27 10.52 .34 .14 IIITMEM50B AF045606.2 1.38 10.07 .46 .20 IIIU2AF1 M96982.1 1.27 12.15 .35 .06 IIIUBASH3A AJ277750.1 1.13 7.32 .17 .21 IIIUBE2G2 AF032456.1 1.17 12.08 .22 .18 IIIW90635 W90635.1 1.09 7.39 .12 .45 IIIWRB BC012415.1 1.21 8.21 .27 .29 IIIZNF295 BC063290.1 1.19 8.78 .25 .27 IIIABCG1 AY048757.1 1.25 8.11 .32 .79 IVADARB1 AY135659.1 1.26 7.28 .34 3.06 IVas-MCM3AP-C21orf85 AW163084.1 1.41 7.71 .50 1.02 IVBRWD1 AB080586.1 1.44 9.14 .53 .38 IVC21orf22 AY040089.1 1.27 7.56 .34 .88 IVC21orf8 AA843704.1 1.09 7.55 .12 1.32 IVCBR1 AB124848.1 1.47 8.20 .55 .76 IVCOL6A1 NM_001848.2 1.60 7.34 .68 .62 IVCSTB AF208234.1 1.35 12.79 .43 .38 IV

(continued)


Table 5. (continued)

Gene Symbol


DCR1-12.0 AJ001868.1 1.28 7.37 .36 .76 IVDCR1-12.0-reverse AJ001868.1 1.44 8.01 .53 1.06 IVDCR1-13.0 AJ001869.1 1.25 9.20 .32 1.26 IVDCR1-13.0-reverse AJ001869.1 1.14 8.98 .19 .91 IVDCR1-15.0 AJ001872.1 1.28 7.08 .36 1.00 IVDSCR4 DQ179113.1 1.32 7.41 .40 .53 IVMX1 AF135187.1 1.49 13.61 .58 .67 IVMX2 M30818.1 1.33 11.94 .41 .48 IVPRDM15 AF426259.1 1.38 8.88 .47 .51 IVPRMT2 (exon 8/9) U80213.1 1.46 8.73 .55 .58 IVTRPM2 AY603182.1 1.08 7.50 .11 1.82 IV

NOTE.—The value A corresponds to for the corresponding[log (DS) � log (control)]/22 2gene across the 40 hybridizations, M corresponds to the mean of log (DS) �2

for the corresponding gene across the 40 hybridizations, is the var-log (control) Var (M)2iance of M, and the DS/control ratio is equal to 2M.

a Class I corresponds to genes expressed proportionally to the gene-dosage effect in DScell lines, class II contains genes that are amplified, class III contains genes that are com-pensated, and class IV contains genes that are highly variable between individuals.

b Oligonucleotide probes mapped to the long isoform of SOD1. See details in the “Discussion”section.

strength of the results, where possible, at least two probesper gene were designed (∼80% of the HSA21 oligoarraycontent). The description of the HSA21 oligoarray contentaccording to BLAST results performed on the latest versionof the human genome sequence (NCBI Gene Databasebuild 36.2) is summarized in table 3. Oligonucleotide se-quences spotted on the array have been designed on thebasis of four main criteria: specificity for the representedsequence, GC content equal to 50%, melting temperatureallowing an optimal match between probe and target atthe hybridization temperature, and no stable predictedsecondary structure.

By use of this new specific high-content HSA21 oli-goarray, 40 differential hybridizations comparing DS andcontrol LCLs were performed. The mean signal intensities(represented in log2 by the A value) of each array spotindicated the expression levels of chromosome 21 genes.A total of 134 genes gave signal intensities above the back-ground cutoff (mean ).A 1 7

Biological Material from Patients with DS and ControlIndividuals

LCLs were obtained after immortalization by EBV of Blymphocytes collected from blood samples of individualswith DS and control individuals. To make sure that EBVtransformation did not induce any chromosomal rear-rangement, all cell lines were karyotyped after immortal-ization. Cell lines were always maintained in exponentialgrowth phase. No significant difference in cell morphol-ogy or cell proliferation was observed between DS andcontrol LCLs.

For three individuals with DS, transcriptome compari-sons between fresh blood samples and LCLs obtained fromthe same individuals were conducted on pangenomic mi-

croarrays.34 From these experiments, no major alterationof the transcriptome by the EBV transformation could bedetected; only 0.5% of the genes exhibited significant dif-ferential expression ( ) (L.D., E.A.Y.-G., and M.-C.P.,P p .01unpublished data).

Experimental Design and Statistical Analysis

The main objective was to detect differentially expressedgenes between DS and control samples, taking into ac-count the sex of and variability between individuals. Theaim of the experimental design was to adapt to the ex-perimental constraints of the study (see table 1 and the“Material and Methods” section). Forty experiments werethus programmed.

First, a mixed model was constructed to highlight theeffects of DS, sex, and DS # sex and to take into accountthe gene-expression variability between individuals. Weused the Benjamini-Hochberg procedure to adjust the Pvalues obtained and to limit false-positive results due tomultiple testing.31 The FDR was set at 0.05. The list ofsignificant genes was thus expected to contain 5% false-positive results.

Chromosome 21 genes did not have significant P valueswhen sex or DS and sex combined (DS # sex) were tested.In other words, chromosome 21 gene expression was notsignificantly different between men and women. In ad-dition, DS effects on gene expression were not dependenton sex. The effects of sex and DS # sex were thus droppedfrom the model, and a simplified mixed model testing theeffects of DS on HSA21 gene expression was ultimatelyused. We first selected genes that had DS/control ratiossignificantly different from 1 (FDR 0.05), using the Ben-jamini-Hochberg procedure.31 Among the 136 expressedtranscripts (134 genes), about half (58) had DS/control


Figure 2. Distribution of DS/control ratios for class I, II, III,and IV genes and non-HSA21 reference genes. The plot representsthe minimum and maximum values (whiskers), the first and thirdquartiles (box), and the median value (midline) of DS/control ratiosfor each class of genes.

ratios different from 1 and always 11, indicating that thesegenes were significantly overexpressed in DS LCLs. Ex-pression ratios of these 58 genes ranged from 1.25 to 2.27,with a mean of 1.5, corresponding to the gene-dosage ef-fect in DS. In parallel, among the 134 expressed genes, weselected those that deviated from this gene-dosage effectwith a ratio significantly different from 1.5. Surprisingly,the majority of expressed transcripts (86) had DS/controlratios significantly different from 1.5 (FDR 0.05). Becauseof this high number, the method described by Storey etal.32 for adjustment of the FDR had to be applied. Resultsshowed that 86 genes had DS/control ratios significantlydifferent from 1.5. Of these 86 genes, 9 had DS/controlratios 11.5, in the range 1.64–2.27, and 77 had DS/controlratios !1.5, in the range 0.74–1.4.

On the basis of this statistical analysis, we classifiedgenes into four categories according to their variation ofexpression between DS and control LCLs, as described intable 4 and represented in figure 1. Class I contained 30genes with DS/control ratios significantly different from1 but not significantly different from 1.5, in the range1.3–1.67. Class II contained nine genes that were signifi-cant in both statistical tests, with DS/control ratios sig-nificantly different from 1, significantly different from 1.5,and 11.5 (range 1.64–2.27). Class III comprised 77 genesthat had DS/control ratios significantly different from 1.5and !1.5 (range 0.74–1.4). The majority of gene predic-tions and antisense transcripts (77%) belonged to thisclass. In addition, gene expression levels of the transcriptsbelonging to class III were significantly lower than thosebelonging to class I ( ) and class II (�5P p 1.32 # 10 P p

). Class IV included the remaining 20 genes with�75.7 # 10

DS/control ratios not significantly different from 1 or from1.5, in the range 1.08–1.6. Table 5 gives the complete listof genes. Distributions (box plots) of DS/control ratios foreach class and for non–chromosome 21 reference genesare shown in figure 2.

The goal of this study was also to demonstrate whetherchromosome 21 gene-expression profiles could differen-tiate DS from control samples. We therefore performedtwo distinct PCAs on the 134 chromosome 21–expressedgenes and the 39 non–chromosome 21 genes used as ref-erences (see the “Material and Methods” section). PCAcould clearly distinguish individuals with DS from controlindividuals, suggesting that the effects of DS on chro-mosome 21 gene expression prevails over any other effect,including biological variability (fig. 3A). In addition, nodistinction could be obtained between individuals withDS and control individuals when PCA was conducted onnon–chromosome 21 genes (fig. 3B). Non–chromosome21 genes had a mean DS/control expression ratio of 1 (fig.2).

Most of the genes present on the HSA21 oligoarray wererepresented by two probes. When the two probes werefound to be expressed, they belonged to the same class,except for C21ORF108 and PRMT2. Concerning C21ORF108,one probe (B1�KIAA0539.eri10102_a) mapping to exon39 of C21ORF108 belonged to class I (DS/control ratio 1.3).The other probe (KIAA0539.gff6561_b), mapping to exon26 of C21ORF108, was in class III (DS/control ratio 1.09).This difference could result from the existence of twoalternative transcripts containing either exon 26 or ex-on 39. Similarly, the two probes representing PRMT2(HRMT1L1.gff2216_a and HRMT1L1.gff2216_b) belongedto classes IV and III, respectively. This difference couldalso be explained by the existence of two alternative tran-scripts containing either exons 5/6 or exons 8/9 describedin the ENSEMBL database.

QPCR Validation Experiments

To confirm variations in gene expression and to validatethe classification of chromosome 21 genes, we performedQPCR on 11 genes belonging to class I (1 gene), class II(2 genes), class III (6 genes), and class IV (2 genes). QPCRwas conducted on all LCLs from individuals with DS andcontrol individuals. Ratios obtained by QPCR confirmedthe classification of chromosome 21 genes deduced fromHSA21 oligoarrays, except for SOD1. SOD1 belonged toclass III and had a ratio of 1.57 by QPCR. However, this1.57-fold difference between LCLs from individuals withDS and control individuals was not significant ( ).P p .15DS/control ratios from QPCR were in agreement with ra-tios obtained from HSA21 oligoarrays (table 6), with a cor-relation coefficient of 0.82. Our HSA21 oligoarray was thusa comprehensive, reproducible, and sensitive tool forstudying gene expression in DS.


Table 6. Comparison between QPCR Results and MicroarrayData

HSA21Gene

GenBankAccessionNumber

Data fromHSA21 Oligoarray

DS/ControlRatio by QPCRa

DS/ControlRatiob Class

SNF1LK AB047786.1 2.14 II 3.36STCH U04735.1 1.97 II 2.06MX1 AF135187.1 1.49 IV 1.78DYRK1A D86550.1 1.41 I 1.77CSTB AF208234.1 1.35 IV 1.49CHAF1B U20980.1 1.29 III 1.38TMEM1 BC101728.1 1.27 III 1.27GART X54199.1 1.17 III 1.38SOD1 AY835629.1 1.15 III 1.57H2BFS AB041017.1 1.13 III 1.05DSCR1 AY325903.1 .93 III 1.11

a The mean expression ratio for the corresponding gene between DSand control cell lines.

b DS/control ratio by QPCR was calculated from normalized Ct obtainedfor DS cell lines relative to control cell lines.

Figure 3. PCA of HSA21 genes (A) and non-HSA21 genes (B). Red and blue symbols represent DS and control samples, respectively.Squares represent samples extracted from females, and diamonds represent samples extracted from males.

Discussion

The aim of the study was to analyze chromosome 21 geneexpression in LCLs from individuals with DS and controlindividuals. Forty differential hybridizations comparingDS LCLs with control LCLs were performed on a dedicatedHSA21 oligoarray designed from the complete humanchromosome 21 gene catalogue (359 genes). Approxi-mately one-third (134) of all chromosome 21 genes, ORFs,and predictions were expressed in LCLs.

On the basis of the expression levels of chromosome 21genes, DS samples were clearly distinct from control sam-ples, thus reflecting the prevalent effect of DS on chro-mosome 21 gene expression. On the contrary, referencegenes mapping to chromosomes other than 21 could notdistinguish DS LCLs from control LCLs. Using the mixed-model analysis, we were able to detect genes that are sig-nificantly overexpressed in DS cell lines (58) and alsogenes that deviate from the gene-dosage effect, with DS/control expression ratios significantly different from 1.5.

Classification of HSA21 Genes

By use of this new data analysis protocol, human chro-mosome 21 genes can now be ranked into four classes bytheir expression levels in DS cell lines relative to controls.This protocol could be applied to expression data obtainedfrom other human tissues, to validate the classification.

Class I contains 30 genes with expression ratio of DS/control close to 1.5 (range 1.3–1.67), correlated to the pres-ence of three genomic copies (table 4 and fig. 1). Theseclass I genes could be responsible for the phenotype ob-served in DS, either directly or indirectly through a sec-ondary effect of cis- or trans-acting genes.

Class II contains nine genes with expression ratio of DS/control 11.64, corresponding to an amplification of the

initial gene dosage (table 4 and fig. 1). Among these genes,SAMSN1, SNF1LK, STCH, and BTG3 show the highest ex-pression ratio, in the range 1.67–2.27.

Gene-dosage amplification could result from a cascad-ing effect through regulation networks involving trans- orcis-acting genes.35 Pellegrini et al.36 identified in silico aputative mitogen-activated kinase cascade with chromo-some 21 kinases involved in various signaling pathways:DYRK1A, SNF1LK, RIPK4, and DSCR3. In our study,DYRK1A, SNF1LK, and DSCR3 were expressed in LCLs,whereas RIPK4 was not. Thus, four replicates were chosenfor each patient.

DYRK1A is under the gene-dosage effect and DSCR3 iscompensated, whereas SNF1LK is amplified from the ini-tial gene dosage. On the basis of this putative mitogen-


Figure 4. Distribution of the variance of M for class I, II, III,and IV genes. The plot represents the minimum and maximumvalues (whiskers), the first and third quartiles (box), and the me-dian value (midline) of the variances of M for each class of genes,where M is the mean of .log (DS) � log (control)2 2

activated kinase cascade, amplification of SNF1LK geneexpression could thus result from the overexpression ofDYRK1A acting as a regulatory factor on SNF1LK in thecascade, and DSCR3 could act as a scaffolding protein.

Class III is the most abundant and contains 77 genes,with a large proportion of gene predictions and antisensetranscripts with DS/control expression ratio !1.4 (table 4and fig. 1). These class III genes are likely to be compen-sated in DS. Compensation mechanisms in trisomic con-ditions have been described in maize and Drosophila37–39

and have been suggested in previous transcriptome stud-ies, both in patients with DS10,12,16 and in mouse mod-els.4,19–21 For example, Lyle et al.20 found that 45% of thetriplicated genes analyzed in their study were compen-sated. Compensation is most likely due to negative feed-backs that would modulate transcriptional activity ormRNA stability of class III genes. Thus, expression of com-pensated genes could be regulated by mechanisms thatare not impaired in DS. For example, trans-inhibitorscouldact directly on the level of expression of these genes. Al-ternatively, trans-activators would activate inhibitors pre-sent in three copies on chromosome 21 and would reducethe expression level of target genes that could be also bepresent on chromosome 21.40 However, the existence ofpolymorph alleles correlated to different levels of expres-

sion should not account for either gene compensation oramplification in a representative population of patientswith DS. A recent study has demonstrated that two CpGislands from human chromosome 21 can be methylatedmonoallelically.41 One of those maps to DSCR3, the otherto C21orf29. Both are class III genes in LCLs.

Six class III genes were tested by QPCR, and all werevalidated, except SOD1. SOD1 is a well-characterized genethat has been shown elsewhere to be overexpressed in DStissues and cells at the RNA and protein levels.13,16,42 InLCLs, the SOD1 gene is transcribed into two variants, along and a short isoform.43 Since SOD1 probes from theHSA21 oligoarray mapped to the long isoform only, theclassification (class III compensated) (table 5) corre-sponded to this long isoform. The ratio deduced from theHSA21 oligoarray (1.15) was found to be slightly lowerthan the one obtained by QPCR for the long isoform(1.57). However, by use of QPCR primers amplifying bothisoforms of SOD1, with the short isoform the most abun-dant in LCLs, we found that the ratio between DS andcontrol LCLs was 1.96 (data not shown). These results sug-gest that, in DS LCLs, SOD1 is overexpressed.

Class IV contains 15 genes and 5 gene predictions thathave DS/control expression ratio not different from either1 or 1.5. These class IV genes are thus highly variablebetween individuals with DS and control individuals. In-deed, figure 4 shows that the variance distribution ofexpression ratios is the highest for class IV genes. Threeclass IV genes (CBR1, PRDM15, and ADARB1) wereshown elsewhere to be highly variable among unaffectedindividuals.44

Using the mixed-model analysis, we have been able todistinguish between gene expression differences resultingfrom DS and those from interindividual variations. In-terindividual variations have been assessed in normalLCLs.45–47 In the present study, we used lymphoblastoidcells established from individuals with DS and control in-dividuals all belonging to Indo-European populations,thus limiting the variations due to ethnic groups.

Copy-number variations have also been described inLCLs.48 They should not have an impact on the resultsunless their frequencies are different in individuals withDS and control individuals, which is unlikely. Moreover,we could not find any correlation between the type ofcopy-number variation (gain or loss) described for partic-ular genes and their gene class. For example, two genesmapping to the same copy-number variant (variation516248) belonged to class II (PDXK) and class IV (CSTB).

Comparison with Expression Data Obtained from DS Tissues

Mao et al.16 studied transcriptome modifications in DSfetal heart, cerebellum, and astrocyte cells, using a pan-genomic Affymetrix U133A chip. Of the 200 genes as-signed to HSA21, 23 were significantly changed in DS tis-sues and 17 were in common with the 58 HSA21 genesthat were significantly changed in our study. Our results


Figure 5. Distribution of expressed, class I, II, III, and IV genes along HSA21. The left Y-axis indicates the proportion of expressedgenes in each 5-Mb interval, and the right Y-axis indicates the proportion of class I, II, III, and IV genes in each 5-Mb interval.

are also in agreement with another gene-expression studyperformed on DS fetal heart cells17 that showed that 16HSA21 genes are significantly overexpressed in fetalhearts. Among these 16 genes, 8 were significantlychanged in our study. Class I, II, and III genes were presentin all tissues, suggesting that gene-dosage effect, amplifi-cation, and compensation are general phenomena andthat LCLs are a good model for studying gene-dosage ef-fects. The differences observed between our study and theothers suggest tissue-specific regulations that have beendescribed elsewhere for the control of GABPa expression.49

Distribution of Gene-Expression Modifications along HSA21

We have analyzed the distribution of expressed genes, aswell as individual gene classes along HSA21. Figure 5shows that expressed genes map preferentially to the distalpart of HSA21, reflecting the nonuniform gene densityalong HSA21q.27 The most telomeric region of HSA21 hasa high proportion of class III genes, perhaps because ofthe presence of a higher proportion of gene predictionsthat are localized inside gene introns.

DS Effects on Alternative Transcripts

To search for differential effects of DS on alternativelyspliced transcripts, we analyzed genes for which oligo-nucleotide probes present on the HSA21 oligoarray coulddifferentiate between alternative transcripts. Seventeengenes had probes specific to alternative transcripts (table7). For seven of those genes, all the probes gave a very

low signal, indicating that these transcripts are not ex-pressed in LCLs. Three genes (C21orf33, C21orf34, andMRPL39) were expressed in LCLs as a unique transcriptand belonged to class I genes, which are overexpressedwith a ratio close to 1.5. For the last seven genes (ADARB1,C21orf66, DYRK1A, GART, PKNOX1, RUNX1, and TMEM1),oligonucleotide probes could distinguish between splicingvariants that had very similar DS/control ratios. Only twoof these genes (C21orf66 and DYRK1A) were significantlyoverexpressed in DS LCLs (i.e., were class I genes), whereasthe others were compensated. These results suggest thatmost of the transcripts belonging to the same gene andexpressed in LCLs are similarly regulated in DS.

DS Effects on Antisense Transcripts

The HSA21 oligoarray was also designed to analyze theeffects of DS on the expression of antisense transcripts.Fourteen antisense transcripts are present with their nest-ing genes on the HSA21 oligoarray (table 8). Among them,10 have been extracted from the HSA21 database estab-lished by Kathleen Gardiner at the Eleanor Roosevelt In-stitute.50 The four remaining antisense transcripts corre-sponded to transcribed sequences in the DCR that havebeen generated from various cDNA mapping and exon-trapping experiments.8,51,52 Seven genes (C21orf25, CHAF1B,DYRK1A, HLCS, KIAA0179, MCM3AP, and TTC3) are ex-pressed in LCLs. Four of the corresponding antisense tran-scripts (as-DYRK1A, as-HLCS, as-KIAA0179, and as-TTC3)were also found to be expressed in LCLs, thus confirming

Table 7. Alternative Transcripts of HSA21 Genes

Gene Symbol and Probe

GenBankAccessionNumber A

DS/ControlRatio M Var(M)

RecognizedVariantsa Class

ADARB1:ADARB1.alt23565_a AY135659.1 5.23 NE NE NE 1, 2, 3, 4 NEADARB1.alt23565_b AY135659.1 4.89 NE NE NE 1, 2, 3, 4 NEADARB1.alt3788_a AY135659.1 7.36 1.22 .29 2.69 1, 2, 4 IVADARB1.alt3788_b AY135659.1 7.21 1.3 .38 3.46 1, 2, 4 IV

C21orf33:HES1.gff1583_b BC003587.1 5.46 NE NE NE 1 NEbi824121 BI824121.1 10.2 1.5 .58 .16 1, 2 IHES1.gff1583_a BC003587.1 10.86 1.53 .62 .25 1, 2 I

C21orf34:orf34�35.eri594_a AF486622.1 5.5 NE NE NE 1 NEC21orf34.gff397_a AF486622.1 5.87 NE NE NE 1, 2 NEC21orf34.gff397_b AF486622.1 5.83 NE NE NE 1, 2 NEorf34�35.alt629_b AF486622.1 6.15 NE NE NE 1, 2 NEorf34�35.eri594_b AF486622.1 5.69 NE NE NE 1, 2 NEC21orf35.gff251_a AF486622.1 5.95 NE NE NE 1, 2, 3 NEorf34�35.alt2559_a AF486622.1 8.15 .95 �.07 .30 1, 2, 3 IIIorf34�35.alt629_a AF486622.1 8.2 .9 �.15 .27 1, 2, 3 III

C21orf66:B3�GCFC.eri2361_a AY033903.1 5 NE NE NE 1, 2, 3 NEB3�GCFC.eri2361_b AY033903.1 5.48 NE NE NE 1, 2, 3, 4 NEGCFC.eri2361_a AY033903.1 10.6 1.56 .64 .09 1, 2, 3 IGCFC.eri2361_b AY033903.1 8.93 1.51 .6 .04 1, 2, 3, 4 IGCFC.gff1083_a AY033903.1 10.19 1.48 .57 .22 1, 2, 3 IGCFC.gff1083_b AY033903.1 8.29 1.47 .56 .12 1, 2, 3 I

DYRK1A:DYRK1.alt2571_a D86550.1 10.33 1.4 .48 .13 1, 2, 3, 4, 5 IDYRK1.alt2571_b D86550.1 10.71 1.4 .49 .22 1, 2, 3, 4, 5 IDYRK1.gff5318_a D86550.1 9.47 1.41 .5 .19 1, 2, 3, 4, 5 IDYRK1.gff5318_b D86550.1 11.02 1.41 .5 .22 1, 2, 3, 4, 5 I

GART:GART.gff3271_a X54199.1 8.79 1.18 .24 .10 1 IIIGART.gff3271_b X54199.1 10.35 1.17 .22 .20 1, 2 III

MRPL39:PRED22.eri707_a AF109357.1 9.46 1.41 .5 .14 1, 2 IPRED22.eri707_b AF109357.1 10.56 1.47 .55 .17 1, 2 IPRED22.gff1072_a AF109357.1 10.33 1.47 .56 .13 1, 2 IPRED22.gff1072_b AF109357.1 10.33 1.46 .55 .18 1, 2 IPRED66.eri187_a AF109357.1 10.28 1.51 .6 .12 1, 2 IPRED66.eri187_b AF109357.1 9.36 1.51 .59 .16 1, 2 IPRED66.gff641_b AF109357.1 8.31 1.42 .5 .17 1, 2 IPRED66.gff641_a AF270511.1 6.58 NE NE NE 2 NE

PKNOX1:PKNOX1.gff3279_a AY196965.1 7.64 1.05 .07 .11 1 IIIPKNOX1.gff3279_b AY196965.1 7.98 1.03 .04 .14 1, 2 III

RUNX1:RUNX1.alt25714_a D43968.1 7.23 .86 �.21 .84 1, 2 IIIRUNX1.alt7267_a D43968.1 7.37 .8 �.32 .74 1, 2 IIIRUNX1.alt7267_b D43968.1 5.56 NE NE NE 2 NERUNX1.gff2722_a D43968.1 7.92 .85 �.23 .74 2 IIIRUNX1.gff2722_b D43968.1 9.03 .89 �.17 .76 2 III

TMEM1:TMEM1.gff5126_a BC101728.1 10.53 1.25 .32 .14 1 IIITMEM1.gff5126_b BC101728.1 10.5 1.28 .36 .14 1, 2 III

NOTE.—The value A corresponds to for the corresponding gene across the 40 hybridizations,[log (DS) � log (control)]/22 2M corresponds to the mean of for the corresponding gene across the 40 hybridizations,log (DS) � log (control)2 2

is the variance of M, and the DS/control ratio is equal to 2M. NE p not expressed.Var (M)a The number of transcript variants hybridizing to the oligonucleotide probe.

Table 8. Sense and Antisense Transcripts on Chromosome 21

Gene Symbol and Probe

GenBankAccessionNumber

IntragenicLocationa A M Var(M)

DS/ControlRatio Class

C21orf56:C21orf56.gff691_a BC084577.1 Exon 2 5.51 NE NE NE NE

as-C21orf56:C21orf56.gff284_a BC084577.1 Exon 4 10.68 .12 .08 1.08 IIIC21orf56.gff284_b BC084577.1 Exon 4 12.79 .08 .08 1.05 III

CHAF1B:CHAF1B.gff2194_a U20980.1 Exon 14 8.66 .37 .12 1.29 IIICHAF1B.gff2194_b U20980.1 Exon 14 8.03 .38 .14 1.30 III

as-CHAF1B:BF740066 BF740066.1 3′ 5.08 NE NE NE NE

HLCS:HLCS.gff6722_a AB063285.1 Exon 12 6.37 NE NE NE NEHLCS.gff6722_b AB063285.1 Exon 11 8.48 .40 .18 1.32 III

as-HLCS:DCR1-8.0_a AJ001862.1 Intron 7 7.58 �.03 .42 .98 IIIDCR1-8.0_b AJ001862.1 Intron 7 5.83 NE NE NE NE

TTC3:TTC3.gff9074_a D84296.1 Exon 47 10.02 .84 .23 1.79 IITTC3.gff9074_b D84296.1 Exon 34 11.17 .83 .17 1.78 II

as-TTC3:bf979681 BF979681.2 Exon 41 9.07 .64 .64 1.56 I

DYRK1A:DYRK1.alt2571_a D86550.1 Exon 13 10.33 .48 .13 1.40 IDYRK1.alt2571_b D86550.1 Exons 7/8 10.71 .49 .22 1.40 IDYRK1.gff5318_a D86550.1 Exon 13 9.47 .50 .11 1.41 IDYRK1.gff5318_b D86550.1 Exon 11 11.02 .50 .22 1.41 I

as-DYRK1A:DCR1-12.0_a AJ001868.1 Intron 1 7.37 .36 .76 1.29 IVDCR1-13.0-RC_a AJ001869.1 Intron 1 8.33 .15 .95 1.11 IVDCR1-13.0-RC_b AJ001869.1 Intron 1 9.55 .22 .91 1.17 IV

KCNJ6:GIRK2(U52153)_a U52153.1 Exon 3 5.60 NE NE NE NEGIRK2(U52153)_b U52153.1 Exon 1 5.28 NE NE NE NE

as-KCNJ6:DCR1-17DCR1-17_a AJ001875.1 Intron 3 7.16 .17 .70 1.12 IV

ADAMTS5:ADAMTS5.gff5523_a AF142099.1 Exon 8 5.39 NE NE NE NEADAMTS5.gff5523_b AF142099.1 Exon 8 5.39 NE NE NE NE

as-ADAMTS5:r18879 R18879.1 Intron 3 5.69 NE NE NE NE

as-C21orf25:aa575913 AA575913.1 3′ 5.85 NE NE NE NE

C21orf25:C21orf25.gff6217_a AB047784.1 Exon 14 8.96 .17 .45 1.13 IIIC21orf25.gff6217_b AB047784.1 Exon 14 8.24 .32 .17 1.25 III

as-CBR3:bi836686 BI836686.1 Exon 3 5.90 NE NE NE NE

CBR3:CBR3.gff878_a AB124847.1 Exon 3 6.18 NE NE NE NECBR3.gff878_b AB124847.1 Exons 1/2 5.31 NE NE NE NE

CLDN14:CLDN14.gff1693_a AF314090.1 Exon 3 6.78 NE NE NE NECLDN14.gff1693_b AP001726.1 3′ 6.05 NE NE NE NE

as-CLDN14:w90592 W90592.1 3′ 5.71 NE NE NE NE

KIAA0179:KIAA0179.gff4984_a D80001.1 Exon 16 5.26 NE NE NE NEKIAA0179.gff4984_b D80001.1 Exon 16 10.15 .28 .24 1.22 III

as-KIAA0179:aa425659 AA425659.1 3′ 8.08 .22 .67 1.17 III

MCM3AP:MCM3.gff6113_a AY590469.1 Exon 27 11.69 .54 .19 1.45 I

MCM3APAS:af426262 AF426262.1 Introns 25–26 5.54 NE NE NE NEaf426263 AF426263.1 Introns 20–21 5.73 NE NE NE NE

NOTE—The value A corresponds to for the corresponding gene across the 40 hybridizations,[log (DS) � log (control)]/22 2M corresponds to the mean of for the corresponding gene across the 40 hybridizations,log (DS) � log (control)2 2

is the variance of M, and the DS/control ratio is equal to 2M. NE p not expressed.Var (M)a The exon or intron to which the oligonucleotide probe maps.


their existence. TTC3 (class II) and its antisense transcript(class I) were overexpressed, whereas the other antisensesequences did not belong to the same class as their cor-responding genes. In addition, two antisense sequences(as-C21orf56 and as-KCNJ6) were expressed in LCLs,whereas their corresponding genes were not. The probereferred to as antisense transcript as-KCNJ6 mapping inintron 3 of KCNJ6, on the opposite strand, correspondsto one of the transcribed sequences isolated in the DCRby exon-trapping experiments.8 Since there is no evidencethat this sequence is an antisense transcript of KCNJ6, itcould thus belong to a gene locus that has not yet beenidentified and might map to the opposite orientation ofKCNJ6.

According to the NCBI Gene Database, C21orf56 (ac-cession number 84221) currently maps on the negativestrand of HSA21 but was previously annotated on the pos-itive strand when the HSA21 oligoarray was designed.Thus, probe sequence representing the antisense tran-script as-C21orf56 could correspond to the actual sensetranscript of C21orf56. Therefore, the expression of anti-sense transcripts is confirmed by our HSA21 oligoarrayexperiments. Sense and antisense transcripts are not al-ways similarly changed in DS.

In conclusion, using our new high-content HSA21 oli-goarrays combined with a new powerful statistical analysisprotocol, we were able to classify HSA21 genes accordingto their level of expression in DS LCLs. We show that,among the expressed transcripts, 29% are sensitive to thegene-dosage effect or are amplified, 56% are compensated,and 15% are highly variable among individuals. Thus,most of the chromosome 21 genes are compensated forthe gene-dosage effect. Gene-expression variations in DSare controlled by mechanisms involving trans and cis reg-ulators acting either directly or through gene-regulationnetworks. Overexpressed genes are likely to be involvedin the DS phenotype, in contrast to the compensatedgenes. Highly variable genes could account for phenotypicvariations observed in patients. Finally, we show that al-ternative transcripts belonging to the same gene are sim-ilarly regulated in DS, whereas sense and antisense tran-scripts are not always similarly regulated. Studies ofhuman tissues by use of the same analysis protocol willvalidate genes that are involved in the DS phenotype.

Acknowledgments

We thank the Banque de Cellules from the Cochin Hospital; Drs.N. Faucon and R. Meloni (CNRS UMR 9923, Hôpital Pitié Sal-petrière, Paris), for advice; D. Leclerre, C. Mikonio, and F. Richard,for their help with spotting the HSA21 oligoarrays; I. Haddad, forthe design of the first version of the HSA21 oligoarray; C. J. Ep-stein (University of California, San Francisco), for very importantcomments on the manuscript; and R. Veitia (Hôpital Cochin,Paris), J.-J. Daudin (Institut National d’Agronomie Paris-Grignon,Paris), Dr. Y. Hérault (CNRS, Orléans), and Dr. P. M. Sinet (InstitutNational de la Santé et de la Recherche Médicale, Paris), for help-ful discussions about the results. E.A.G. had a fellowship from

the Ministère de la Recherche, and G.G had a fellowship fromthe Fondation Jérôme Lejeune. This work was supported by theFondation Jérôme Lejeune, European Economic Communitygrant T21 Targets, and AnEUpolidy.

Web Resources

Accession numbers and URLs for data presented herein are asfollows:

Eleanor Roosevelt Institute: Chromosome 21 Gene Function andPathway Database, http://chr21db.cudenver.edu/

GenBank, http://www.ncbi.nlm.nih.gov/Genbank/ (for accessionnumbers in tables 5–8)

Gene Expression Omnibus (GEO), http://www.ncbi.nlm.nih.gov/geo/ (for accession number GSE6408)

Max Planck Institute: Chromosome 21 Gene Catalog Based onthe New AGP File July 2002, http://chr21.molgen.mpg.de/chr21_catalogs/chr21_mar_2002.html

NCBI Entrez, http://www.ncbi.nlm.nih.gov/gquery/gquery.fcgi(for accession numbers L13852 and AB000468)

NCBI Gene Database, http://www.ncbi.nlm.nih.gov/sites/entrez(for accession number 84221)

Online Mendelian Inheritance in Man (OMIM), http://www.ncbi.nlm.nih.gov/Omim/ (for DS)

The R Project for Statistical Computing, http://www.r-project.org/

References

1. Epstein CJ, Korenberg JR, Anneren G, Antonarakis SE, AymeS, Courchesne E, Epstein LB, Fowler A, Groner Y, Huret JL, etal (1991) Protocols to establish genotype-phenotype corre-lations in Down syndrome. Am J Hum Genet 49:207–235

2. Antonarakis SE, Lyle R, Dermitzakis ET, Reymond A, DeutschS (2004) Chromosome 21 and Down syndrome: from gen-omics to pathophysiology. Nat Rev Genet 5:725–738

3. Lejeune J, Gautier M, Turpin R (1959) Etudes des chromo-somes somatiques de neuf enfants mongoliens. C R HebdSeances Acad Sci 248:1721–1722

4. Saran NG, Pletcher MT, Natale JE, Cheng Y, Reeves RH (2003)Global disruption of the cerebellar transcriptome in a Downsyndrome mouse model. Hum Mol Genet 12:2013–2019

5. Shapiro BL (2001) Developmental instability of the cerebel-lum and its relevance to Down syndrome. J Neural TransmSuppl 61:11–34

6. Schinzel A (2001) Catalogue of unbalanced chromosome ab-errations in man. Walter de Gruyter, Berlin

7. Olson LE, Roper RJ, Sengstaken CL, Peterson EA, Aquino V,Galdzicki Z, Siarey R, Pletnikov M, Moran TH, Reeves RH(2007) Trisomy for the Down syndrome “critical region” isnecessary but not sufficient for brain phenotypes of trisomicmice. Hum Mol Genet 16:774–782

8. Dahmane N, Ghezala GA, Gosset P, Chamoun Z, Dufresne-Zacharia MC, Lopes C, Rabatel N, Gassanova-Maugenre S,Chettouh Z, Abramowski V, et al (1998) Transcriptional mapof the 2.5-Mb CBR-ERG region of chromosome 21 involvedin Down syndrome. Genomics 48:12–23

9. Delabar J, Theophile D, Rahmani Z, Chettouh Z, Blouin J,Prieur M, Noel B, Sinet P (1993) Molecular mapping oftwenty-four features of Down syndrome on chromosome 21.Eur J Hum Genet 1:114–124

10. FitzPatrick DR, Ramsay J, McGill NI, Shade M, Carothers AD,


Hastie ND (2002) Transcriptome analysis of human autoso-mal trisomy. Hum Mol Genet 11:3249–3256

11. Gross SJ, Ferreira JC, Morrow B, Dar P, Funke B, Khabele D,Merkatz I (2002) Gene expression profile of trisomy 21 pla-centas: a potential approach for designing noninvasive tech-niques of prenatal diagnosis. Am J Obstet Gynecol 187:457–462

12. Mao R, Zielke CL, Zielke HR, Pevsner J (2003) Global up-regulation of chromosome 21 gene expression in the devel-oping Down syndrome brain. Genomics 81:457–467

13. Giannone S, Strippoli P, Vitale L, Casadei R, Canaider S, LenziL, D’Addabbo P, Frabetti F, Facchin F, Farina A, et al (2004)Gene expression profile analysis in human T lymphocytesfrom patients with Down syndrome. Ann Hum Genet 68:546–554

14. Tang Y, Schapiro MB, Franz DN, Patterson BJ, Hickey FJ,Schorry EK, Hopkin RJ, Wylie M, Narayan T, Glauser TA, etal (2004) Blood expression profiles for tuberous sclerosis com-plex 2, neurofibromatosis type 1, and Down’s syndrome. AnnNeurol 56:808–814

15. Chung IH, Lee SH, Lee KW, Park SH, Cha KY, Kim YS, Lee S(2005) Gene expression analysis of cultured amniotic fluidcell with Down syndrome by DNA microarray. J Korean MedSci 20:82–87

16. Mao R, Wang X, Spitznagel E, Frelin L, Ting J, Ding H, KimJW, Ruczinski I, Downey T, Pevsner J (2005) Primary and sec-ondary transcriptional effects in the developing humanDown syndrome brain and heart. Genome Biol 6:R107

17. Li CM, Guo M, Salas M, Schupf N, Silverman W, Zigman W,Husain S, Warburton D, Thaker H, Tycko B (2006) Cell type-specific over-expression of chromosome 21 genes in fibro-blasts and fetal hearts with trisomy 21. BMC Med Genet 7:24

18. Amano K, Sago H, Uchikawa C, Suzuki T, Kotliarova SE, Nu-kina N, Epstein CJ, Yamakawa K (2004) Dosage-dependentover-expression of genes in the trisomic region of Ts1Cjemouse model for Down syndrome. Hum Mol Genet 13:1333–1340

19. Kahlem P, Sultan M, Herwig R, Steinfath M, Balzereit D, Ep-pens B, Saran NG, Pletcher MT, South ST, Stetten G, et al(2004) Transcript level alterations reflect gene dosage effectsacross multiple tissues in a mouse model of Down syndrome.Genome Res 14:1258–1267

20. Lyle R, Gehrig C, Neergaard-Henrichsen C, Deutsch S, An-tonarakis SE (2004) Gene expression from the aneuploidchromosome in a trisomy mouse model of Down syndrome.Genome Res 14:1268–1274

21. Dauphinot L, Lyle R, Rivals I, Dang MT, Moldrich RX, GolfierG, Ettwiller L, Toyama K, Rossier J, Personnaz L, et al (2005)The cerebellar transcriptome during postnatal developmentof the Ts1Cje mouse, a segmental trisomy model for Downsyndrome. Hum Mol Genet 14:373–384

22. Gardiner K (2006) Transcriptional dysregulation in Downsyndrome: predictions for altered protein complex stoichi-ometries and post-translational modifications, and conse-quences for learning/behavior genes ELK, CREB, and the es-trogen and glucocorticoid receptors. Behav Genet 36:439–453

23. Wester U, Bondeson ML, Edeby C, Anneren G (2006) Clinicaland molecular characterization of individuals with 18p de-letion: a genotype-phenotype correlation. Am J Med GenetA 140:1164–1171

24. Hattori M, Fujiyama A, Taylor TD, Watanabe H, Yada T, Park

HS, Toyoda A, Ishii K, Totoki Y, Choi DK, et al (2000) TheDNA sequence of human chromosome 21. Nature 405:311–319

25. Kapranov P, Cawley SE, Drenkow J, Bekiranov S, StrausbergRL, Fodor SPA, Gingeras TR (2002) Large-scale transcriptionalactivity in chromosomes 21 and 22. Science 296:916–919

26. Reymond A, Camargo AA, Deutsch S, Stevenson BJ, Parmi-giani RB, Ucla C, Bettoni F, Rossier C, Lyle R, Guipponi M(2002) Nineteen additional unpredicted transcripts from hu-man chromosome 21. Genomics 79:824–832

27. Gardiner K, Fortna A, Bechtel L, Davisson MT (2003) Mousemodels of Down syndrome: how useful can they be? Com-parison of the gene content of human chromosome 21 withorthologous mouse genomic regions. Gene 318:137–147

28. Quackenbush J (2002) Microarray data normalization andtransformation. Nat Genet Suppl 32:496–501

29. Wolfinger RD, Gibson G, Wolfinger ED, Bennett L, HamadehH, Bushel P, Afshari C, Paules RS (2001) Assessing gene sig-nificance from cDNA microarray expression data via mixedmodels. J Comput Biol 8:625–637

30. SAS Institute (2000) SAS/STAT software release 8. Cary, NC31. Benjamini Y, Hochberg Y (1995) Controlling the false dis-

covery rate: a practical and powerful approach to multipletesting. J Roy Stat Soc B 57:289–300

32. Storey JD, Tibshirani R (2003) Statistical significance for ge-nomewide studies. Proc Natl Acad Sci USA 100:9440–9445

33. Janel N, Christophe O, Ait Yahya-Graison E, Hamelet J, PalyE, Prieur M, Delezoide AL, Delabar JM (2006) Paraoxonase-1expression is up-regulated in Down syndrome fetal liver.Biochem Biophys Res Commun 346:1303–1306

34. Brigand KL, Russell R, Moreilhon C, Rouillard JM, Jost B,Amiot F, Magnone V, Bole-Feysot C, Rostagno P, Virolle V, etal (2006) An open-access long oligonucleotide microarray re-source for analysis of the human and mouse transcriptomes.Nucleic Acids Res 34:e87

35. Potier MC, Rivals I, Mercier G, Ettwiller L, Moldrich RX, Laf-faire J, Personnaz L, Rossier J, Dauphinot L (2006) Transcrip-tional disruptions in Down syndrome: a case study in theTs1Cje mouse cerebellum during post-natal development. JNeurochem Suppl 1 97:104–109

36. Pellegrini-Calace M, Tramontano A (2006) Identification of anovel putative mitogen-activated kinase cascade on humanchromosome 21 by computational approaches. Bioinformat-ics 22:775–778

37. Birchler JA (1979) A study of enzyme activities in a dosageseries of the long arm of chromosome one in maize. Genetics92:1211–1229

38. Devlin RH, Holm DG, Grigliatti TA (1982) Autosomal dosagecompensation in Drosophila melanogaster strains trisomic forthe left arm of chromosome 2. Proc Natl Acad Sci USA 79:1200–1204

39. Guo M, Birchler JA (1994) Trans-acting dosage effects on theexpression of model gene systems in maize aneuploids. Sci-ence 266:1999–2002

40. Birchler JA, Riddle NC, Auger DL, Veitia RA (2005) Dosagebalance in gene regulation: biological implications. TrendsGenet 21:219–226

41. Yamada Y, Watanabe H, Miura F, Soejima H, Uchiyama M,Iwasaka T, Mukai T, Sakaki Y, Ito T (2004) A comprehensiveanalysis of allelic methylation status of CpG islands on hu-man chromosome 21q. Genome Res 14:247–266

42. Sinet PM, Lavelle F, Michelson AM, Jerome H (1975) Super-


oxide dismutase activities of blood platelets in trisomy 21.Biochem Biophys Res Commun 67:904–909

43. Sherman L, Levanon D, Lieman-Hurwitz J, Dafni N, GronerY (1984) Human Cu/Zn superoxide dismutase gene: molec-ular characterization of its two mRNA species. Nucleic AcidsRes 12:9349–9365

44. Deutsch S, Lyle R, Dermitzakis ET, Attar H, Subrahmanyan L,Gehrig C, Parand L, Gagnebin M, Rougemont J, JongeneelCV, et al (2005) Gene expression variation and expressionquantitative trait mapping of human chromosome 21 genes.Hum Mol Genet 14:3741–3749

45. Cheung VG, Conlin LK, Weber TM, Arcaro M, Jen KY, MorleyM, Spielman RS (2003) Natural variation in human gene ex-pression assessed in lymphoblastoid cells. Nat Genet 33:422–425

46. Spielman RS, Bastone LA, Burdick JT, Morley M, Ewens WJ,Cheung VG (2007) Common genetic variants account fordifferences in gene expression among ethnic groups. Nat Ge-net 39:226–231

47. Storey JD, Madeoy J, Strout JL, Wurfel M, Ronald J, Key JM

(2007) Gene-expression variation within and among humanpopulations. Am J Hum Genet 80:502–509

48. Redon R, Ishikawa S, Fitch KR, Feuk L, Perry GH, AndrewsTD, Fiegler H, Shapero MH, Carson AR, Chen W, et al (2006)Global variation in copy number in the human genome. Na-ture 444:444–454

49. O’Leary DA, Pritchard MA, Xu D, Kola I, Hertzog PJ, RistevskiS (2004) Tissue-specific overexpression of the HSA21 geneGABPa: implications for DS. Biochim Biophys Acta 1739:81–87

50. Nikolaienko O, Nguyen C, Crinc LS, Cios KJ, Gardiner K(2005) Human chromosome 21/Down syndrome gene func-tion and pathway database. Gene 364:90–98

51. Chen H, Chrast R, Rossier C, Morris MA, Lalioti MD, Anton-arakis SE (1996) Cloning of 559 potential exons of genes ofhuman chromosome 21 by exon trapping. Genome Res 6:747–760

52. Ohira M, Seki N, Nagase T, Suzuki E, Nomura N, Ohara O,Hattori M, Sakaki Y, Eki T, Murakami Y, et al (1997) Geneidentification in 1.6-Mb region of the Down syndrome regionon chromosome 21. Genome Res 7:47–58

Classiffication of Human Chromosome 21 Gene-Expression Variations in Down Syndrome: Impact on Disease PhenotypesMaterial and MethodsCell Lines and Culture ConditionsHuman Chromosome 21 (HSA21) OligoarrayExperimental DesignmRNA Extraction, HSA21 Oligoarray Hybridization, Data Filtering, and NormalizationExpression Data Analysis: Statistical TestingPCR Experiments

ResultsDesign of a Comprehensive HSA21 OligoarrayBiological Material from Patients with DS and Control IndividualsExperimental Design and Statistical AnalysisQPCR Validation ExperimentsClassification of HSA21 GenesComparison with Expression Data Obtained from DS TissuesDistribution of Gene-Expression Modi.cations along HSA21DS Effects on Alternative TranscriptsDS Effects on Antisense Transcripts

AcknowledgmentsReferences

classiﬁcation of human chromosome 21 gene-expression ... · down syndrome caused by chromosome 21...

Documents